• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제15권10호
• /
• pp.3608-3626
• /
• 2021

The detection of bolts is an important task in high-speed train inspection systems, and it is frequently performed to ensure the safety of trains. The difficulty of the vision-based bolt inspection system lies in small sample defect detection, which makes the end-to-end network ineffective. In this paper, the problem is resolved in two stages, which includes the detection network and cascaded classification networks. For small bolt detection, all bolts including defective bolts and normal bolts are put together for conducting annotation training, a new loss function and a new boundingbox selection based on the smallest axis-aligned convex set are proposed. These allow YOLOv3 network to obtain the accurate position and bounding box of the various bolts. The average precision has been greatly improved on PASCAL VOC, MS COCO and actual data set. After that, the Siamese network is employed for estimating the status of the bolts. Using the convolutional Siamese network, we are able to get strong results on few-shot classification. Extensive experiments and comparisons on actual data set show that the system outperforms state-of-the-art algorithms in bolt inspection.

• Smart Structures and Systems
• /
• 제21권5호
• /
• pp.687-694
• /
• 2018

The problem of wheel tread defects has become a major challenge for the health management of high-speed rail as a wheel defect with small radius deviation may suffice to give rise to severe damage on both the train bogie components and the track structure when a train runs at high speeds. It is thus highly desirable to detect the defects soon after their occurrences and then conduct wheel turning for the defective wheelsets. Online wheel condition monitoring using wheel impact load detector (WILD) can be an effective solution, since it can assess the wheel condition and detect potential defects during train passage. This study aims to develop an FBG-based track-side wheel condition monitoring method for the detection of wheel tread defects. The track-side sensing system uses two FBG strain gauge arrays mounted on the rail foot, measuring the dynamic strains of the paired rails excited by passing wheelsets. Each FBG array has a length of about 3 m, slightly longer than the wheel circumference to ensure a full coverage for the detection of any potential defect on the tread. A defect detection algorithm is developed for using the online-monitored rail responses to identify the potential wheel tread defects. This algorithm consists of three steps: 1) strain data pre-processing by using a data smoothing technique to remove the trends; 2) diagnosis of novel responses by outlier analysis for the normalized data; and 3) local defect identification by a refined analysis on the novel responses extracted in Step 2. To verify the proposed method, a field test was conducted using a test train incorporating defective wheels. The train ran at different speeds on an instrumented track with the purpose of wheel condition monitoring. By using the proposed method to process the monitoring data, all the defects were identified and the results agreed well with those from the static inspection of the wheelsets in the depot. A comparison is also drawn for the detection accuracy under different running speeds of the test train, and the results show that the proposed method can achieve a satisfactory accuracy in wheel defect detection when the train runs at a speed higher than 30 kph. Some minor defects with a depth of 0.05 mm~0.06 mm are also successfully detected.

• 한국음향학회지
• /
• 제28권5호
• /
• pp.416-423
• /
• 2009

16개 압전요소로 구성된 3 MHz의 선형 위상차배열 초음파 트랜스듀서를 대상으로, 모든 압전요소가 정상일 때와 임의의 요소 하나가 결함으로 인해 동작하지 않을 때의 음장을 이론적으로 시뮬레이션하고, 슈리렌법에 따라 구축한 음장가시화장치를 이용하여 실험적으로 측정하였다. 조향각 $0^{\circ}$와 $30^{\circ}$일 때 각각에 대한 시뮬레이션의 결과, 임의의 압전요소가 결함으로 인해 동작하지 않을 때의 음장은 모든 요소가 정상적으로 동작할 때의 음장에 비해 부엽 패턴이 크게 다르게 나타나며, 그 형태는 가시화에 의한 측정결과와 잘 일치하였다. 따라서, 가시화 장치에 의해 측정된 2차원 음장에서의 부엽패턴을 시뮬레이션 결과와 비교 분석함으로써 선형 위상차배열 초음파 트랜스듀서의 결함요소 검출이 가능함을 알았다.

• 비파괴검사학회지
• /
• 제34권4호
• /
• pp.305-310
• /
• 2014

본 고에서는 미세 이물질이 삽입된 볼베어링에 대하여 비파괴평가를 제안하였다. 비파괴평가 연구로서 동적인 하중조건하 회전체의 동작에 따른 고장 진단을 위해 비접촉식 정량화된 적외선 열화상 기법을 적용하였다. 이로부터 볼베어링에 대한 적정 체결조건을 설정하였고 고장 상태감시에 대한 수동형 열화상시험을 수행하였다. 본 연구로부터, 적외선 열화상 시험은 조기의 결함 진단을 평가하기 위해 정상 및 이물질이 삽입된 시편들로부터의 온도 프로파일링을 비교, 분석되었다. 연구의 비파괴검사 평가의 결과로써, 고장에 이르는 이상단계에 따른 볼베어링의 온도 특성이 정량적으로 분석되었다.

• 센서학회지
• /
• 제28권6호
• /
• pp.371-376
• /
• 2019

We have developed an acoustic resonance inspection system to inspect the welding defects in the mechanical parts fabricated using friction stir welding method. The inspection system was consisted of a DAQ board, a microphone sensor, an impact hammer, and controlled by a PC software. The system was developed to collect and analyze the sound signal generated by hitting the sample with an impact hammer to determine whether it is defective. In this study, 100% welded good samples were compared with 95%, 90%, and 85% welded samples, respectively. The variation of the completeness in welding did not affect the visual appearance in the samples. As a result of analyzing the natural frequencies of the good samples, the five natural frequency peaks were identified. In the case of the defective samples, the frequency change was observed. The welding failure detection time was fast enough to be only 0.7 seconds. Employing our welding defect inspection system to the actual industrial field will maximize the efficiency of quality inspection and thus improve the productivity.

• International Journal of Computer Science & Network Security
• /
• 제22권10호
• /
• pp.1-10
• /
• 2022

Nowadays software defect prediction (SDP) is most active research going on in software engineering. Early detection of defects lowers the cost of the software and also improves reliability. Machine learning techniques are widely used to create SDP models based on programming measures. The majority of defect prediction models in the literature have problems with class imbalance and high dimensionality. In this paper, we proposed Centroid and Nearest Neighbor based Class Imbalance Reduction (CNNCIR) technique that considers dataset distribution characteristics to generate symmetry between defective and non-defective records in imbalanced datasets. The proposed approach is compared with SMOTE (Synthetic Minority Oversampling Technique). The high-dimensionality problem is addressed using Ant Colony Optimization (ACO) technique by choosing relevant features. We used nine different classifiers to analyze six open-source software defect datasets from the PROMISE repository and seven performance measures are used to evaluate them. The results of the proposed CNNCIR method with ACO based feature selection reveals that it outperforms SMOTE in the majority of cases.

• 한국전자통신학회논문지
• /
• 제15권1호
• /
• pp.101-106
• /
• 2020

본 연구는 다양한 어셈블리 부품의 용접부 내부결함을 검사하기 위한 초음파 탐상 장비 개발에 관한 것이다. 본 연구에서는 초음파 탐상을 위하여 시스템의 모션제어 S/W, 초음파 송수신기 제어, 결함 판정 기준 설정 등의 계측 S/W 등이 설계되었으며, 양품과 불량품의 비교분석을 하기 위하여 용접결함 불량품 샘플워크 등도 제작되었다. 이와 같은 구성으로 이루어진 시스템을 통하여 어셈블리 부품 용접부의 결함 위치 및 깊이에 대한 자동검사 기능을 수행할 수 있었으며, 종전에 전문가에 의해 이루어졌던 용접부의 내부결함에 대한 판단을 시스템이 수행하도록 하였다.

• 한국표면공학회지
• /
• 제55권6호
• /
• pp.398-402
• /
• 2022

Vacuum Insulation Panel (VIP) is an high energy efficient insulation system that facilitate slim but high insulation performance, based on based on a porous core material evacuated and encapsulated in a multi-barrier envelope. Although VIP has been on the market for decades now, it wasn't until recently that efforts have been initiated to propose a standard on aging testing. One of the issues regarding VIP is its durability and aging due to pressure and moisture dependent increase of the initial low thermal conductivity with time. It is hard to visually determine at an early stage. Recently, a method of analyzing the damage on the a material surface by applying image processing technology has been widely used. These techniques provide fast and accurate data with a non-destructive way. In this study, the surface VIP images were analyzed using the Harris corner detection algorithm. As a result, 171,333 corner points in the normal packaging were detected, whereas 32,895 of the defective packaging, which were less than the normal packaging. were detected. These results are considered to provide meaningful information for the determination of VIP condition.

• 산업경영시스템학회지
• /
• 제45권4호
• /
• pp.86-98
• /
• 2022

Recently, many studies have been conducted to improve quality by applying machine learning models to semiconductor manufacturing process data. However, in the semiconductor manufacturing process, the ratio of good products is much higher than that of defective products, so the problem of data imbalance is serious in terms of machine learning. In addition, since the number of features of data used in machine learning is very large, it is very important to perform machine learning by extracting only important features from among them to increase accuracy and utilization. This study proposes an anomaly detection methodology that can learn excellently despite data imbalance and high-dimensional characteristics of semiconductor process data. The anomaly detection methodology applies the LIME algorithm after applying the SMOTE method and the RFECV method. The proposed methodology analyzes the classification result of the anomaly classification model, detects the cause of the anomaly, and derives a semiconductor process requiring action. The proposed methodology confirmed applicability and feasibility through application of cases.

• 한국정보통신학회:학술대회논문집
• /
• 한국정보통신학회 2022년도 춘계학술대회
• /
• pp.108-111
• /
• 2022

제품의 결함 탐지를 위한 머신 비전 시스템에 딥러닝을 적용하기 위해서는 다양한 결함 사례에 대한 방대한 학습 데이터가 필요하다. 하지만 실제 제조 산업에서는 결함의 종류에 따른 데이터 불균형이 생기기 때문에 결함 사례를 일반화할 수 있을 만큼의 제품 이미지를 수집하기 위해서는 많은 시간이 소요된다. 본 논문에서는 적은 데이터로도 학습이 가능한 샴 신경망을 제품 결함 탐지에 적용하고, 제품 결함 이미지 데이터의 속성을 고려하여 이미지 쌍 구성법과 대조 손실 함수를 수정하였다. AUC-ROC로 샴 신경망의 임베딩 성능을 간접적으로 확인한 결과, 같은 제품끼리만 쌍을 구성하고 결함이 있는 제품 간에는 쌍을 구성하였을 때, 그리고 지수 대조 손실로 학습하였을 때 좋은 임베딩 성능을 보였다.